Walk into any radio club and float the idea of running single sideband through a class C amplifier, and someone will tell you flatly that it cannot be done. The class C stage, biased so hard that it conducts for only a sliver of each radio frequency cycle, is the textbook villain of linear amplification. It chops the waveform, throws away half the information, and smears the spectrum across the band. For decades this has hardened into a rule of thumb that brooks no argument: SSB demands a linear amplifier, full stop, and class C is for CW and FM where amplitude carries nothing. The rule is mostly right, which is exactly why the exceptions are so instructive. The interesting question is not whether class C distorts an SSB envelope, because it obviously does, but whether the impossibility is a property of the amplifier or of how the signal is presented to it.
The answer turns out to hinge on a distinction that the blanket prohibition glosses over. A class C stage cannot amplify an amplitude-varying signal that is already complete. But amplitude information does not have to arrive at the final already impressed on the carrier. If the envelope is stripped off and reintroduced at the final itself, the class C stage becomes not a linearity problem but the heart of an efficient transmitter. The myth of impossibility is really a myth about where in the chain the modulation lives.
Why a class C stage destroys an envelope that arrives intact
Start with the mechanism, stated precisely. An amplifier's conduction angle is the fraction of each radio frequency cycle during which the device conducts current. A class A stage conducts the full 360 degrees, class B exactly 180 degrees, class AB somewhere between, and class C strictly less than 180 degrees, often far less. The device sits biased beyond cutoff, and only the peaks of the drive that exceed the cutoff threshold produce any output at all.
That threshold is the whole trouble for an amplitude-modulated input. An SSB envelope rides up and down in amplitude, carrying the voice. When the instantaneous envelope is large, the drive comfortably exceeds the cutoff threshold and the stage conducts. When the envelope dips low, as it does between syllables and on quiet passages, the drive never reaches the threshold and the stage produces nothing. The transfer relationship between input envelope and output envelope is not a straight line but a broken one, flat at zero below the threshold and curved above it. Express the output amplitude Vout against input amplitude Vin and the relation looks like
Vout = 0 for Vin < Vth
Vout = k * (Vin - Vth) for Vin > Vth (roughly, near threshold)
That offset Vth and the curvature near it are precisely the nonlinearity that generates intermodulation. A two-tone test reveals it instantly. Drive the stage with two equal tones and the broken transfer curve mixes them, producing third order products at frequencies 2f1 - f2 and 2f2 - f1 that fall right inside the passband. The amplitude of the third order product relative to the tones grows steeply with drive, and for a hard class C stage those products sit only a handful of decibels down, a spectral mess that earns immediate complaints from the neighbors. The tuned output tank restores a clean sine wave at the carrier frequency, which is why class C works beautifully for CW and FM, but the tank cannot restore the envelope information that the cutoff threshold already discarded.
The conduction angle and the efficiency it buys
The reason anyone tolerates this brutal distortion is efficiency, and the numbers explain why the temptation never goes away. The theoretical maximum drain or plate efficiency of an amplifier climbs as the conduction angle shrinks. For an idealized stage the maximum efficiency runs roughly as follows across the classes:
class A: 50 percent maximum
class B: 78.5 percent maximum
class C: typically 70 to 80 percent and rising toward higher values as conduction angle falls
A linear stage suitable for SSB, operating in class A or class B, dissipates a large fraction of its input power as heat. The patent literature puts hard numbers on it: a linear amplifier preserving the modulation envelope typically runs at 20 to 30 percent efficiency, so producing 3 watts of clean output demands about 15 watts from the supply and dumps roughly 12 watts as heat. A class C stage at 40 to 80 percent efficiency would deliver the same 3 watts from a fraction of the input power. That gap, a factor of two or three in wasted power, is the prize that keeps engineers circling back to the supposedly impossible combination.
The efficiency advantage is not marginal. In a battery-powered rig it is the difference between an hour of operation and three. In a high-power station it is the difference between a final that needs forced-air cooling and one that runs on a modest heat sink. The blanket prohibition asks the builder to walk away from that prize, and the more sophisticated approaches refuse to.
Stripping the envelope off and putting it back at the final
The escape from the trap is a technique with a long pedigree, known as envelope elimination and restoration, devised by Kahn in the early days of SSB. The insight is that an SSB signal can be decomposed into two separate streams, a constant-amplitude phase-modulated carrier and a separate amplitude envelope, and these two can be handled by completely different hardware before being recombined at the output.
The signal is split at low level. A limiter strips all amplitude variation from the SSB signal, leaving a constant-envelope wave that carries only the phase and frequency information. That constant-amplitude wave has no amplitude information for a class C stage to destroy, so it can be amplified by the most efficient class C or class D stage available with no linearity concern whatsoever. Meanwhile, an envelope detector recovers the amplitude variation as a baseband signal. The recovered envelope is then reimposed on the amplified carrier by modulating the supply voltage of the final stage, exactly the high-level modulation principle used in classic plate-modulated AM transmitters.
The mathematics of the recombination is clean. The output is the product of the constant-amplitude phase-carrying wave and the restored envelope:
s_out(t) = A(t) cos(2pifct + phi(t))
where A(t) is the restored envelope applied through the supply and phi(t) is the phase carried by the limited wave. The final stage runs in efficient class C the entire time, conducting hard, while the amplitude information arrives not through the radio frequency drive but through the modulated supply rail. The modulation occurs in the final, not before it, which is the exact condition the old CB-radio builders described when they noted that their class C amplifiers handled AM because the modulation happened inside the final stage rather than upstream.
Why the recombination must track in both amplitude and time
The technique is not free, and its difficulties are where the real engineering lives. The first is bandwidth matching between the two paths. The envelope path, running through a supply modulator, is inherently slower than the phase path, running at radio frequency. If the envelope arrives at the final even slightly delayed relative to the phase, the product A(t) and cos term no longer align, and the misalignment generates distortion of its own. The two paths must be time-aligned to a small fraction of the envelope period, which for voice means matching delays to tens of microseconds and for wider signals far tighter.
The second difficulty is the linearity of the supply modulator itself. The envelope restoration is only as clean as the modulator that applies it. Any nonlinearity in the supply modulator maps directly onto the output envelope, so the linearity problem has not vanished, it has merely moved from the radio frequency final to the supply modulator, where it is easier to solve because it operates at baseband. A switching modulator handling the envelope can itself be highly efficient, preserving the overall efficiency gain, but it must be designed with the same care a linear designer would lavish on the final.
The third subtlety is what happens at envelope nulls. An SSB envelope occasionally passes through zero, and at that instant the supply voltage to the class C final drops toward zero too. The phase is changing most rapidly exactly at these nulls, demanding the widest bandwidth from both paths precisely where the signal level is lowest and the tracking hardest. This is why a naive implementation sounds rough on sharp consonants and why practical designs add correction to handle the nulls gracefully.
Closing the loop to police the residual distortion
A further refinement wraps feedback around the whole arrangement to catch the distortion the open-loop scheme leaves behind. The patent literature on band-limited modulators using class C amplification describes exactly this: closed-loop feedback around the amplifier assures linear modulation over a wide range and preserves the frequency characteristics of the baseband signal on the modulation envelope. The loop senses the actual output envelope, compares it to the intended envelope, and corrects the supply modulation to drive the error toward zero.
The feedback also does double duty as a filter. By incorporating the closed-loop system response into the overall filter characteristic, the design adds an effective filter pole that helps reject modulation harmonics, so the loop cleans the spectrum and shapes the band in one move. The result is a transmitter that runs its final in efficient class C, restores the envelope through a modulated and feedback-corrected supply, and emerges with a spectrum clean enough for SSB service. The supposed impossibility has been engineered around not by making class C linear, which cannot be done, but by arranging the signal so that class C never has to be linear in the first place.
A numerical comparison of the two transmitter architectures
Numbers make the trade unmistakable, so follow the same 100 watt SSB output through both architectures. Take a linear class B final at 30 percent average efficiency on voice, a realistic figure once the duty cycle of speech is folded in, since the envelope spends much of its time well below peak where class B efficiency sags. The DC input power is
Pdc = Pout / eta = 100 / 0.30 = 333 W
and the heat dissipated is
Pdiss = Pdc - Pout = 333 - 100 = 233 W
That 233 watts must leave through the heat sink, demanding a large thermal mass and often forced air.
Now the envelope elimination and restoration path. The class C final runs at, say, 75 percent efficiency, and the supply modulator that restores the envelope runs at 90 percent because it switches rather than dissipates. The two efficiencies multiply, because the envelope power passes through the modulator before reaching the final:
eta_total = eta_final eta_modulator = 0.75 0.90 = 0.675
so the DC input for the same 100 watt output is
Pdc = 100 / 0.675 = 148 W
and the dissipation is
Pdiss = 148 - 100 = 48 W
The comparison is stark. The linear final wastes 233 watts; the restored class C final wastes 48 watts, a factor of nearly five less heat for identical output. The supply current tells the same story. On a 50 volt rail the linear final draws
I = 333 / 50 = 6.7 A
against the class C architecture's
I = 148 / 50 = 3.0 A
less than half the battery drain. This single calculation, 48 watts of waste against 233, is why the supposedly impossible combination keeps tempting designers and why the architectures that make it work are worth the trouble of building. The cost is complexity, two signal paths that must track in time and amplitude, but the reward is a transmitter that runs cool and sips current while putting out a clean sideband.
What the myth gets right and what it hides
The blanket statement that class C cannot amplify SSB is a useful lie. It is correct for the naive case, the case a beginner is most likely to attempt, where a complete SSB signal is fed into the radio frequency input of a hard-biased stage. In that case the cutoff threshold shreds the envelope, the two-tone test lights up with intermodulation, and the neighbors reach for the phone. Teaching the rule as absolute protects beginners from a predictable failure.
But the rule hides a richer truth that the experienced builder should hold. Linearity is not a property the final stage must possess; it is a property the overall transmitter must achieve. A class C final embedded in an envelope elimination and restoration architecture, with a linear or feedback-corrected supply modulator and careful time alignment between the phase and amplitude paths, delivers clean SSB at efficiencies a linear amplifier cannot touch. The modulation simply has to happen in the right place, at the final through the supply, rather than upstream through the drive. Seen this way, the question was never whether class C can carry a sideband. It was always where the amplitude information should live, and the moment that question is asked correctly, the impossibility dissolves into an engineering problem with several good solutions.
Our Telegram Channel
Our YouTube Channel